There’s an expression “Deus ex machina” (the god from the machine) that describes a plot where things are resolved by the timely arrival (as, with some stage device) of some unexpected character or agent–such as an authority figure–to settle things that have gotten out of hand.  (Like, the arrival of the police at the end of Monty Python and the Holy Grail–how else were they going to be able to end it?)  Here we look at something rather different–making 3d printing machines from 3d printing machines, or machinae ex machinis. [fixed ablative plural–thx Andreas]

One of the motivating concepts behind reprapping is a device that makes itself.  A big recursion.  In one sense, this is an evolutionary design hypothesis.  In another, it is a test of machine fabrication robustness.   In the case of the Mendel printer,  we have a device that can make at least the plastic parts for itself.   That’s nice.  The Mendel, however, is still dependent on other resources for its electronics,  its metal parts, and its assembly.   It’s a start.  A good start.  But it’s only a bit of the way.

If we broaden the idea out a bit, we can see how local manufacturing can develop a capability to produce new things–even 3d printers–using a composite approach.   The machine doesn’t have to make itself entirely for things to be interesting.   In Investigations, Stuart Kauffman argues that cross-dependent chemical systems may be actually more stable and easier for the universe to get to than auto-dependent ones.  That is, it is easier to develop two systems that each supply the other with inputs than to have one system that feeds itself.

In 3d printing, we can use powder printers to create molds to generate the plastic parts for RepRap printers.   For some, this breaks with their idea of the printer making itself.   Our suggestion:  run with the idea a bit.  It’s a big world, with lots of things to do yet.  Using powder 3d printers to help build lots of plastic 3d printers is an expectable evolutionary development–two printers capable of making parts for each other, and doing it better than a single printer can do it, at least for now.   One might imagine, even, a third device that assembles both powder and plastic printers from the outputs of the other two.  Then we would have something of a technology ecology, the beginnings of a  Cradle to Cradle technology cycle for 3d printing.   While machine self-production is one way to do it, why exclude machine symbiosis, with fabricating machines looking out for each other?

As we explore casting–using 3d printers to make the templates and/or the molds–for various parts, and in a wide range of materials–we expand the range of available capabilities for local manufacturing.   This is a good thing, too, though it may not be an immediate goal of the RepRap project.  We can 3d print the part, or 3d print a mold for the part, or 3d print the positive from which to make the mold.  In any of these, the local 3d print capability is the enabling technology, and each has a useful place.   Whether it takes us one step or three, we are still creating parts locally, and there’s still an assembly step, whether it is casting the parts and drilling holes or  going to the store for threaded rod and nuts and getting out the hacksaw.

From an evolutionary point of view, machine symbiosis and casting is more efficient than a single machine slaving for a week over the parts for an alter ego.   It’s the evolutionarily competitive thing to do.  As 3d printing capabilities get better–faster, especially–casting, or any other process,  may not to be thing anymore.  As the recent article in The Economist notes, 3d printing plastic parts is approaching the price point for injection molding.

It’s possible that we could take hybrid 3d printing further.  What if folks made molds for the metal parts of a Mendel or Prusa?   Then we could be casting those parts in our material of choice.  How about structural support in cast glass or bronze using 3d printed molds?  or 3d printed wood?   Would this be a bad thing?  No way.  It’s still not really a machine that makes itself, but if it’s a machine that a human wants to make, and can make, then that’s also a good thing.

Working this way, we imagine manufactured artifacts composed of different classes of parts that are variously self-made,  generic, and specialized.   A Prusa is a composite of plastic parts (self-made), metal parts (off the shelf generics), and electronics (specially sourced and rather specialized).  Similarly, we can see a bicycle or cell phone or sofa this way.   As these undergo redesigns, some parts will become much easier to produce locally via 3d printing, or via casting from 3d printed molds (or similar things done for textiles).  Rather than it being all or none (that is, all parts made in a factory, or all parts self-made by a single 3d printer), we point to a new kind of craft engagement with broader and more immediate goals–to democratize fabrication, to broaden the available materials and modalities, and to integrate into existing resources where these provide advantages.

Of course, there is still the big driver of a connection to a computer of some form that provides the software platform to drive the printer.  That’s a big dependency, to be sure, with a whole industry involved in creating computers.  It makes the self-produced 3d printer look more than a bit dependent.  That is, the machine has the capability for making its plastic parts, but only if it is hooked up to a computer running software that manages it.   It’s only an output device that is making part of itself, not the whole device.

Back in the day, before software, before even the ENIAC moment, there were cards to control machines, even player pianos had paper punch rolls.  It’s lost technology, now, limited to things like music boxes.   I wonder if a new electro-mechanical control mechanism might be developed in parallel with the software/electronic one, that would allow for an off-line 3d printer to be able to make an object without having to boot a computer as controller.  Perhaps a 3d printer could make all its own plastic parts without software control.  That would be a different sort of project.  A printer able to print plastic cards (or filaments) with the commands to create its own parts.

Our broad effort is to increase the capabilities of humans to manage their own design and fabrication.  There is a new fabrication shop coming, and the tools are bringing together solid modeling, CNC milling, additive manufacturing, and smart assembly.   As various contributors add their pieces we get closer to what this shop will look like.  It’s pretty neat.  There will be machines from machines, and maybe someday, one machine robust enough to make all its own parts.  We aren’t there yet, but we can help to pick up the evolutionary pace.

5 Comments on Machinae ex machinis

  1. Andreas says:

    Great article! But I think the title should be “Machinae ex machinis” to be correct Latin 😉

  2. barnett says:

    yup, blew my plural 1st declension case, fixed now, I think. I hate it with classical languages–no strong type checking at compile. thx!

  3. […] the parts. Still not convinced? How about this: the molds can be created by a 3D printer or by using a high-resolution power printing method like they have […]

  4. Leo Dearden says:


    The idea of a manufacturing ecosystem is nicely put. I think (as you seem to) that we should shift the focus of our attention from building self-manufacturing individual machines, to building self-sustaining manufacturing ecosystems.

    We already have one manufacturing ecosystem that is, at least in the short term, self sustaining: The global one. It’s not robust though, and it’s rather bloated and generally inelegant. 🙂

    I think that the quickest and best route to a better manufacturing ecosystem (distributed, freedom enhancing, clean, resilient, etc) is to use the current system as a bootstrap. If we accept it for the moment and replace its functions piece-wise and as appropriate, then we get the maximum rate of evolution.

    The open source ecology project ( is the most complete and furthest developed part of this process that I know of.

    • barnett says:


      Nice points. Yes, consider a set of machines creating interesting and useful things, including parts for each other. That may be evolutionarily more competitive than a system committing parthenogenesis. It is clear that we are moving toward new ways of thinking about fabrication, design, and assembly. The open source ecology project looks like a good early indicator of what’s possible.

Leave a Reply